Greetings, 

The two groups can be compared through a factorial analysis, or if the raw data are available, a combined ANOVA can utilize the both group in addition to their interaction under one condition that is the data must contain replications. Otherwise, you may choose a confounded factorial design or so.

thank you Hamzah

but it's not about comparing two groups. a factorial analysis is already done and for explaining one of the interactions the author has performed another separate Anova in each group. I wonder if it's a correct post hoc since the powers and p values are compared.

Best

If I understood it correctly, the author performed a (2?) factorial ANOVA, found a significant interaction and used two separate one factorial ANOVA to evaluate the effect within each group? This is not wrong, but this approach has less power as for example a simple effects analysis following a significant interaction.

BUT something seems to be odd in your description, because in your original question you state "to clarify a significant interaction between these two groups" which implies that you are interested in the comparison of this two groups, where the  approach to test each group seperately would not answer the question. In your answer you say " it's not about comparing two groups". Ok, what then? Within or between the groups? If you want to answer the question, how big is the effect within each group you could compare (partial) eta square, which will not answer if the two groups differ significantely.

Please clarify your query.

Are only comparable, if the experimental design is the same. This does not mean that they have made the same ANOVA.

Hi

thanks for your answers.

Dear Rainer, sorry for bad explanation. the two ANOVA were performed to clarify the interaction between group and time (pre and posttest) while the difference between groups were not significant. the author aimed to see in which group, the method was more effective. therefore your first assumption is true.